Question about submitting node with Apache Arrow dependency

Questions
1

Hello n8n team,

We’ve developed a custom node for n8n:

:backhand_index_pointing_right: @faim-group/n8n-nodes-faim

The node provides serverless time-series forecasting via our FAIM API.

Technically, our backend expects and returns data in the Apache Arrow IPC binary format (Content-Type: application/vnd.apache.arrow.stream), which allows very compact transfer of large tabular and time-series data.

:gear: The issue

We can safely remove all other dependencies — but Apache Arrow (apache-arrow) is essential for:

  • Serializing/deserializing Arrow IPC streams

  • Handling schema and field metadata

  • Efficiently transmitting columnar data (float64 arrays, timestamps, etc.)

Replacing Arrow with JSON or CSV would significantly degrade performance, since the payloads would grow by several megabytes per request.

:red_question_mark:Question

Is it possible to submit an n8n node that includes the apache-arrow dependency, or is there any approved workaround for heavy scientific/data packages like this?

If Arrow itself isn’t permitted, could you please suggest:

  • Whether it can be bundled or whitelisted

  • Or if there’s a recommended way to handle binary IPC streams efficiently within n8n nodes

Thank you for your help!

We’d love to make this node available publicly, as it brings large-model forecasting capabilities directly into n8n workflows.

Best regards,

Andrei Chernov

FAIM

faim.it.com

2

Hi @Andrei_Chernov

Recommended approach

  • Run Arrow-based processing in an external service (Python/Node/Rust)

  • Expose it via HTTP, gRPC, or file storage

  • n8n workflows handle orchestration, sending references and receiving results

  • Use n8n’s binary support or pass URLs for large datasets instead of raw IPC streams

n8n is best used as an orchestrator; heavy scientific/data computation should stay outside the node.